Automated Detection of Retrogressive Thaw Slumps in the High Arctic Using High-Resolution Satellite Imagery
Retrogressive thaw slumps (RTS) are considered one of the most dynamic permafrost disturbance features in the Arctic. Sub-meter resolution multispectral imagery acquired by very high spatial resolution (VHSR) commercial satellite sensors offer unique capacities in capturing the morphological dynamic...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-08-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/14/17/4132 |
| _version_ | 1797493472471547904 |
|---|---|
| author | Chandi Witharana Mahendra R. Udawalpola Anna K. Liljedahl Melissa K. Ward Jones Benjamin M. Jones Amit Hasan Durga Joshi Elias Manos |
| author_facet | Chandi Witharana Mahendra R. Udawalpola Anna K. Liljedahl Melissa K. Ward Jones Benjamin M. Jones Amit Hasan Durga Joshi Elias Manos |
| author_sort | Chandi Witharana |
| collection | DOAJ |
| description | Retrogressive thaw slumps (RTS) are considered one of the most dynamic permafrost disturbance features in the Arctic. Sub-meter resolution multispectral imagery acquired by very high spatial resolution (VHSR) commercial satellite sensors offer unique capacities in capturing the morphological dynamics of RTSs. The central goal of this study is to develop a deep learning convolutional neural net (CNN) model (a UNet-based workflow) to automatically detect and characterize RTSs from VHSR imagery. We aimed to understand: (1) the optimal combination of input image tile size (array size) and the CNN network input size (resizing factor/spatial resolution) and (2) the interoperability of the trained UNet models across heterogeneous study sites based on a limited set of training samples. Hand annotation of RTS samples, CNN model training and testing, and interoperability analyses were based on two study areas from high-Arctic Canada: (1) Banks Island and (2) Axel Heiberg Island and Ellesmere Island. Our experimental results revealed the potential impact of image tile size and the resizing factor on the detection accuracies of the UNet model. The results from the model transferability analysis elucidate the effects on the UNet model due the variability (e.g., shape, color, and texture) associated with the RTS training samples. Overall, study findings highlight several key factors that we should consider when operationalizing CNN-based RTS mapping over large geographical extents. |
| first_indexed | 2024-03-10T01:20:28Z |
| format | Article |
| id | doaj.art-e677cc35a9ed43499a6f782fbdc11b8c |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T01:20:28Z |
| publishDate | 2022-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-e677cc35a9ed43499a6f782fbdc11b8c2023-11-23T14:01:26ZengMDPI AGRemote Sensing2072-42922022-08-011417413210.3390/rs14174132Automated Detection of Retrogressive Thaw Slumps in the High Arctic Using High-Resolution Satellite ImageryChandi Witharana0Mahendra R. Udawalpola1Anna K. Liljedahl2Melissa K. Ward Jones3Benjamin M. Jones4Amit Hasan5Durga Joshi6Elias Manos7Department of Natural Resources and the Environment, University of Connecticut, Storrs, CT 06269, USADepartment of Natural Resources and the Environment, University of Connecticut, Storrs, CT 06269, USAWoodwell Climate Research Center, Falmouth, MA 02540, USAInstitute of Northern Engineering, University of Alaska Fairbanks, Fairbanks, AK 99775, USAInstitute of Northern Engineering, University of Alaska Fairbanks, Fairbanks, AK 99775, USADepartment of Natural Resources and the Environment, University of Connecticut, Storrs, CT 06269, USADepartment of Natural Resources and the Environment, University of Connecticut, Storrs, CT 06269, USADepartment of Natural Resources and the Environment, University of Connecticut, Storrs, CT 06269, USARetrogressive thaw slumps (RTS) are considered one of the most dynamic permafrost disturbance features in the Arctic. Sub-meter resolution multispectral imagery acquired by very high spatial resolution (VHSR) commercial satellite sensors offer unique capacities in capturing the morphological dynamics of RTSs. The central goal of this study is to develop a deep learning convolutional neural net (CNN) model (a UNet-based workflow) to automatically detect and characterize RTSs from VHSR imagery. We aimed to understand: (1) the optimal combination of input image tile size (array size) and the CNN network input size (resizing factor/spatial resolution) and (2) the interoperability of the trained UNet models across heterogeneous study sites based on a limited set of training samples. Hand annotation of RTS samples, CNN model training and testing, and interoperability analyses were based on two study areas from high-Arctic Canada: (1) Banks Island and (2) Axel Heiberg Island and Ellesmere Island. Our experimental results revealed the potential impact of image tile size and the resizing factor on the detection accuracies of the UNet model. The results from the model transferability analysis elucidate the effects on the UNet model due the variability (e.g., shape, color, and texture) associated with the RTS training samples. Overall, study findings highlight several key factors that we should consider when operationalizing CNN-based RTS mapping over large geographical extents.https://www.mdpi.com/2072-4292/14/17/4132Arcticpermafrostretrogressive thaw slumpsatellite imagesdeep learning |
| spellingShingle | Chandi Witharana Mahendra R. Udawalpola Anna K. Liljedahl Melissa K. Ward Jones Benjamin M. Jones Amit Hasan Durga Joshi Elias Manos Automated Detection of Retrogressive Thaw Slumps in the High Arctic Using High-Resolution Satellite Imagery Remote Sensing Arctic permafrost retrogressive thaw slump satellite images deep learning |
| title | Automated Detection of Retrogressive Thaw Slumps in the High Arctic Using High-Resolution Satellite Imagery |
| title_full | Automated Detection of Retrogressive Thaw Slumps in the High Arctic Using High-Resolution Satellite Imagery |
| title_fullStr | Automated Detection of Retrogressive Thaw Slumps in the High Arctic Using High-Resolution Satellite Imagery |
| title_full_unstemmed | Automated Detection of Retrogressive Thaw Slumps in the High Arctic Using High-Resolution Satellite Imagery |
| title_short | Automated Detection of Retrogressive Thaw Slumps in the High Arctic Using High-Resolution Satellite Imagery |
| title_sort | automated detection of retrogressive thaw slumps in the high arctic using high resolution satellite imagery |
| topic | Arctic permafrost retrogressive thaw slump satellite images deep learning |
| url | https://www.mdpi.com/2072-4292/14/17/4132 |
| work_keys_str_mv | AT chandiwitharana automateddetectionofretrogressivethawslumpsinthehigharcticusinghighresolutionsatelliteimagery AT mahendrarudawalpola automateddetectionofretrogressivethawslumpsinthehigharcticusinghighresolutionsatelliteimagery AT annakliljedahl automateddetectionofretrogressivethawslumpsinthehigharcticusinghighresolutionsatelliteimagery AT melissakwardjones automateddetectionofretrogressivethawslumpsinthehigharcticusinghighresolutionsatelliteimagery AT benjaminmjones automateddetectionofretrogressivethawslumpsinthehigharcticusinghighresolutionsatelliteimagery AT amithasan automateddetectionofretrogressivethawslumpsinthehigharcticusinghighresolutionsatelliteimagery AT durgajoshi automateddetectionofretrogressivethawslumpsinthehigharcticusinghighresolutionsatelliteimagery AT eliasmanos automateddetectionofretrogressivethawslumpsinthehigharcticusinghighresolutionsatelliteimagery |